Databases: Introduction and overview. Akhil Kumar Smeal College of Business Penn State University

Transcription

1 Databases: Introduction and overview Akhil Kumar Smeal College of Business Penn State University

2 Outline Examples/Motivation Data modeling Transaction management Next generation databases Conclusions

3 Favorite application Travel he ultimate "Aloha" Experience: Sabre Reservation system FEATURES: Travel services from 400 airlines, 60,000 hotels and 50 car rental companies, railroads, cruise lines; Builds one billion fares, updated 5 times per day; Handles bookings for 35% of all reservations made world-wide amounting to more than $70 Billion annually. Average transaction time: less than 3 seconds, Peak transaction rate: transactions per second. Annual bookings: more than 300 million. Equipment: 31 mainframes,57 terabytes of storage. Processing Power: 11,000 MIPS of processing power. Security System: withstand high winds, tornadoes, earthquakes Protection: 40 inches of reinforced concrete System reliability: %

4 Favorite application - Banking

5 Favorite application at a bar A bar database Swipe driver s license at bar Look up name, address, birth date, etc. Summarize information On Tuesdays over-40 crowd is large Thursday customers come from upscale zip codes Women make up 80% of crowd when Elvis performs

6 Data Model Consists of concepts and tools for describing Basic or raw data ("price of a chair is 50.99") Data relationships ("ABC Co. supplies chairs") Data semantics ("Unit of price is US $") Data constraints ("A Minimum shipment is 1000")

7 Entity Relationship Modeling Build an E-R model of real world situations Entities (objects or things of interest) E.g. faculty, students, courses, offerings Relationships between entities E.g. Erica is taking Database from Prof. Ball Used widely for database design build E-R model and then convert to the relational model

8 E-R Diagram for manufacturing company [Aside: In a hotel database, does a reservation belong to guest, room, date?]

9 ER Modeling Example and Basic Symbols Entity Type symbol Relationship symbol Entity Type name Primary Key Attributes Course CourseNo CrsDesc CrsUnits Has Relationship name Offerings SectionNo Location Time An entity is an important "thing" in the application area. A relationship relates two or more entities together.

10 Cardinality Notation Single line: one cardinality Inside symbol: minimum cardinality Crow's foot: many cardinality Course CourseNo CrsDesc CrsUnits Outside symbol: maximum cardinality Has Circle: zero cardinality Offering SectionNo Location Time

11 The Three-Schema Architecture view1 view2 view3 External Schema (User View) Conceptual Schema Physical Schema

12 Three Schema Architecture for Database Development External Schema Describes a part of a database that is of interest to a particular user (also called a view ). Subset of conceptual schema Conceptual Schema Describes structure of the entire database (e.g., ER model, relational model) Independent of a specific DBMS No details of physical design Physical Schema (or Internal schema) Describes physical structure of the database (tables, indexes, access methods, database distribution.) Conceptual schema is independent of physical schema - data independence.

13 Database Development Phases Data requirements Conceptual Data Modeling ERD Logical Database Design Tables Physical Database Design Internal Schema

14 Logical data model - Relational Model In the relational model, all data is stored in relations or tables. Example Relations (or Tables) The above relations may be described as follows: SALES1(Cust_ID, Name, Salesperson) SPERSON(Salesperson, Region)

15 Relational Model Attributes Example of tabular data in the relational model Customer-id customername customerstreet customercity accountnumber Johnson Alma Palo Alto A Smith North Rye A Johnson Alma Palo Alto A Jones Main Harrison A Smith North Rye A-201

16 A Sample Relational Database

17 Petstore Database - ER diagram Supplier SupplierID Name ContactName Phone Address ZipCode CityID AnimalOrder OrderID OrderDate ReceiveDate SupplierID ShippingCost EmployeeID City CityID ZipCode City State AreaCode Population1990 Population1980 Country Latitude Longitude Merchandise Order PONumber OrderDate ReceiveDate SupplierID EmployeeID ShippingCost Animal OrderItem OrderID AnimalID Cost Employee EmployeeID LastName FirstName Phone Address ZipCode CityID TaxPayerID DateHired DateReleased OrderItem PONumber ItemID Quantity Cost Breed Category Breed Category Category Registration Animal AnimalID Name Category Breed DateBorn Gender Registered Color ListPrice Photo Sale SaleID SaleDate EmployeeID CustomerID SalesTax Merchandise ItemID Description QuantityOnHand ListPrice Category SaleAnimal SaleID AnimalID SalePrice Customer CustomerID Phone FirstName LastName Address ZipCode CityID SaleItem SaleID ItemID Quantity SalePrice This is a simplified, high-level ER diagram showing one-to-one and one-to-many relationships. Primary keys are in bold, minimum cardinalities are not shown.

18 Example Queries on Petstore Database List all animals with yellow in their color. List all dogs with yellow in their color born after 6/1/01. List all merchandise for cats with a list price greater than $10. List all dogs who are male and registered or who were born before 6/1/01 and are grey in color. What is the average sale price of all animals? What is the total cost we paid for all animals? List the top 10 customers and total amount they spent. How many cats are in the animal list? Count the number of animals in each category. List the CustomerID of everyone who bought something between 4/1/01 and 5/31/01. List the first name and phone of every customer who bought something between 4/1/01 and 5/31/01. List the last name and phone of anyone who bought a registered white cat between 6/1/01 and 12/31/01. Which employee has sold the most items?

19 Simple SQL Queries List all columns for all animals SELECT * FROM Animal; List all columns for all animals whose color is yellow. SELECT * FROM Animal WHERE (Color LIKE '%yellow%'); List only AnimalId, Category, Breed and Color columns for all animals whose color contains the letters yellow. SELECT AnimalID, Category, Breed, Color FROM Animal WHERE (Color LIKE %yellow% ); The % is a wildcard for any string of characters. The text inside single quotes is case-sensitive.

20 Models Relational model Entity-Relationship model Other models: Object-oriented models (such as UML) Semi-structured data models XML Based models (Older models: hierarchical model and network model)

21 Transaction Management Concurrency control and crash recovery Transactions must be interleaved, i.e., concurrent A database without a concurrency control mechanism will be compromised due to interference between users: Lost update problem, caused by write-write conflict Dirty read, caused by read-write conflict A real world databases must allow proper concurrent access

22 Lost Update Problem - Airlines Transaction A Time Transaction B Read SR (10) T 1 T 2 Read SR (10) If SR > 0 then T 3 SR = SR -1 T 4 If SR > 0 then SR = SR -1 Write SR (9) T 5 T 6 Write SR (9) SR: Seats Remaining

23 Lost Update Problem: Banking Husband and wife to take money from the same account with $1000 Husband H Wife W read $1000 READ $1000 withdraw $100 Withdraw $200 write $900 WRITE $800 Possible sequence without concurrency control 1. H read $ W READ $ H write $ W WRITE $800 Balance is $900 instead of the correct value of $700. One update is lost.

24 Uncommitted Dependency or Dirty Read Problem Transaction A Time Transaction B Read SR (10) T 1 SR = SR - 1 T 2 Write SR (9) T 3 T 4 Read SR (9) Rollback T 5 SR: Seats Remaining

25 Locking Fundamentals Locking is the main tool of concurrency control Rule: Obtain lock before accessing a data item and release lock at end of transaction Wait if a conflicting lock is held Shared lock (S-Lock): conflicts with exclusive locks Exclusive lock (X-Lock): conflicts with all other kinds of locks Concurrency control manager maintains the lock table

26 Crash recovery Must keep a log: When Ti writes an object: the old value and the new value. Log record transferred to disk before the changed page! Ti commits/aborts: a log record indicating this action. Log records are chained together by Transaction id, so it s easy to undo a specific Transaction Log is often duplexed and archived on stable storage. [Ti: is a transaction running in the database]

27 Next generation databases Features: Sources of data are widespread and their number is potentially infinite The sources vary in nature from being unstructured text on the Web to semi-structured data such as articles, reports and news briefs, to structured sources like conventional database systems Application areas may vary from business data to scientific (astronomy, biology, etc.) and geospatial data Data may also vary across the media spectrum (from plain text, to sound, graphics and pictures) and also across languages

28 New applications interesting problems Digitize 50 years of BBC coverage and serve on demand Random access to 1 petabyte of data (10 18 bytes) GIS applications ("alert me if there is cancellation for a show/game and I am within 1 mile) Analyzing click stream data, continuous stream data Voice to natural language to structured data Bio-informatics (analyzing large biological data sets) Storing large numbers of gene sequences Finding gene locations and querying gene sequences Learn structure and functions of proteins

29 Conclusions Database systems have come a long way from humble beginnings as simple file systems Discussed basic principles of database systems Mentioned lots of potential new applications